| Issue |
171920
|
| Summary |
[missed-opt] [isel] Unnecessary shift count masking in 128-bit arithmetic
|
| Labels |
new issue
|
| Assignees |
|
| Reporter |
purplesyringa
|
[Godbolt](https://godbolt.org/z/WeWG3x1TM)
```cpp
__uint128_t shift(unsigned long x, unsigned long n) {
return (__uint128_t)x << (n & 63);
}
```
```asm
; x86-64
shift(unsigned long, unsigned long):
mov rcx, rsi
mov rax, rdi
shl rax, cl
shr rdi
and cl, 63 ; unnecessary
not cl
shr rdi, cl
mov rdx, rdi
ret
```
```asm
; aarch64
shift(unsigned long, unsigned long):
and x8, x1, #0x3f ; unnecessary
lsr x9, x0, #1
lsl x0, x0, x1
eor x8, x8, #0x3f
lsr x1, x9, x8
ret
```
![]()
_______________________________________________
llvm-bugs mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-bugs
